[FIXED] Saving an image after tf.image.resize

Issue

I am trying to resize some images using tensorflows image resizing method. I have a loop where I load in the image, resize and then try to write it back out using cv2.imwrite(). However, it continues to error out, does this not output an image? Here's my code:

j=0
for i in images:
  skyr_img = imread(my_data_dir+'/'+str(i)) #converts image to  an array
  skyr_img=tf.image.resize(skyr_img,[171,256],antialias=True,method='bilinear')
  print(skyr_img.shape,j)
  cv2.imwrite(my_data_dir+'/train/img'+str(j),skyr_img)
  j=j+1

and this is the error I get:

(171, 256, 4) 0
---------------------------------------------------------------------------
error                                     Traceback (most recent call last)
<ipython-input-49-80ea3aae6498> in <module>
      4   skyr_img=tf.image.resize(skyr_img,[171,256],antialias=True,method='bilinear')
      5   print(skyr_img.shape,j)
----> 6   cv2.imwrite(my_data_dir+'/train/img'+str(j),skyr_img)
      7   j=j+1

error: OpenCV(4.6.0) :-1: error: (-5:Bad argument) in function 'imwrite'
> Overload resolution failed:
>  - img is not a numpy array, neither a scalar
>  - Expected Ptr<cv::UMat> for argument 'img'

Solution

If you are using colab you can mount your drive to colab using

from google.colab import drive
drive.mount('/content/drive', force_remount=True)

You can save images to your drive using

img = cv2.imread('/content/gold.jpeg')
resized_img = tf.image.resize(img, (1050, 1610)).numpy()
path ='/content/drive/MyDrive'
cv2.imwrite(os.path.join(path , 'fish.jpg'), resized_img)

Please refer to this working gist. Thank You.

Answered By - TFer